Propagate callable overrides to ideate deduplication and avoid duplicate response_fn/get_all_responses_fn in whatever

[hemanth-asirvatham]

Motivation

• Ensure custom callables (response_fn / get_all_responses_fn) are consistently forwarded into the ideation deduplication step so user-provided overrides are honored.
• Prevent passing duplicate response_fn/get_all_responses_fn entries into nested runner invocations which caused "multiple values for keyword argument" errors.

Description

• Added a new deduplicate_run_kwargs parameter to the top-level ideate API and included it in the _with_callable_overrides plumbing so response_fn/get_all_responses_fn propagate into the ideation deduplication run (updated src/gabriel/api.py).
• Forwarded deduplicate_run_kwargs to Ideate.run so nested deduplication receives those run-time overrides (updated call-site in ideate).
• Modified the whatever helper to consume any response_fn / get_all_responses_fn from **kwargs when the explicit parameters are not provided, and to remove them from kwargs if the explicit parameters are set, preventing duplicate keyword passing to nested runner.run (updated src/gabriel/api.py).
• Updated ideate docstring to mention deduplicate overrides in the list of fine-grained _run_kwargs.

Testing

• No automated tests were executed as part of this change; changes are limited to argument plumbing in src/gabriel/api.py and require integration/runtime verification with user callables.

---

Codex Task

[github-actions]

Thank you for your submission, we really appreciate it. Like many open-source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution. You can sign the CLA by just posting a Pull Request Comment same as the below format.

---

I have read the CLA Document and I hereby sign the CLA

---

_{You can retrigger this bot by commenting recheck in this Pull Request. Posted by the CLA Assistant Lite bot.}